With Nukissiorfiit’s mission to be 100% fossil-fuel independent and its need to add more hydropower plant infrastructure to achieve its goal, accurate and more robust financial forecasting became a critical need for the utility.

Nukissiorfiit had worked with CogniTech’s consultants since 2015 as part of another organization. The company trusted the consultants’ advice that to meet its aggressive goals, Nukissiorfiit could benefit from an AI-infused planning and analytics solution.

In 2018, CogniTech invited a delegation of five people from the utility to a week-long IBM Analytics conference in Stockholm. The Nukissiorfiit delegation was the largest contingent attending from a single company. The intent was to get inspired for new digital solutions, be more data-driven and talk about new versions of IBM® Cognos® Analytics and IBM Planning Analytics, as well as AI solutions. While discussing another project, Andersen-Aagaard became interested in learning more about the planning project.

“We needed to move beyond traditional budgeting and start working dynamically with our forecasting, using monthly rolling forecasts with a longer horizon of 18 months rolling, to continuously incorporate the newest information,” says Andersen-Aagaard. “And we knew that the new process should not lead to a disproportionate burden where 70 people spend time on it every month. We could see that with the more than 300 sub-budgets we were handling, we couldn’t expand this process and continue to use the traditional way. So we needed to think differently, and that’s where the AI-driven rolling forecasts come into play.”

Kai Erik Ettrup, Partner at CogniTech, explains: “We pitched the idea of using AI and machine learning to improve the process, and Nukissiorfiit bought into the idea.” The utility emphasized that the new tool would have to have a good user interface with an efficient input and verification process. “The inspirational sessions we had in Stockholm gave us the confidence to move ahead with the platform and the collaboration with CogniTech,” says Andersen-Aagaard.

CogniTech met with Nukissiorfiit in workshops to discuss how the company could migrate from its old planning and forecasting solution to a rolling forecast with machine learning and predictive forecasting built in.

CogniTech then helped the utility migrate to the Cognos Analytics solution, the AI-driven business intelligence platform that supports the analytics cycle, from discovery to operationalization, and provides a governed approach to manage, explore and visualize data. Its AI assistant allows the user to interact in exactly the places that need attention and dismiss the 80% of overviews and financial data that don’t need attention.

CogniTech’s Moeller Madsen explains that one of Nukissiorfiit’s main goals in migrating to the rolling forecasts solution from IBM was to enable the utility to make better business decisions.

Andersen-Aagaard, agreeing, adds: “You have a small window in Greenland where you can make outdoor repairs or initiate a new infrastructure project or develop a new power plant. It’s very important that you have 100% visibility and the right insights to make a decision on whether we have the cash flow to do it. For example, if our people in the south of the country have a delayed project portfolio, are the expenses then coming in later, or is it delayed sufficiently that we can push ahead with other projects in the north? This can be quite a daunting planning task to manage when you have 100 - 200 larger infrastructure projects concurrently running.”

Nukissiorfiit relies on the Cognos Analytics solution to gather input from company employees and transform it into a set of reports that top management uses to make planning decisions about its continuous work with infrastructure projects. The company would otherwise have to be much more cautious in approving all the new infrastructure grids, new power plants and substations, new water plants and new renewable energy solutions.

“We had a lot of considerations concerning how the solution could be architected to provide maximum credibility,” Andersen-Aagaard says. “Our financial books consist of a lot of areas, including turnover, variable costs, capacity costs such as salary, depreciation, interest costs, and financial costs. All of these areas have unique properties. We were faced with the classical dilemma in budgeting that it’s difficult for people to incorporate all factors. A machine can handle this much better if historical data is extrapolated into the future.” The company has already set up more than 100 different reports that take different aspects of its financial information and provide it to the managers for them to act on.

“With the help of CogniTech, we were able to eliminate a tremendous amount of administrative work for our technical employees, who hardly took much joy in the task,” Andersen-Aagaard says. “In addition, they could confirm results from the AI as opposed to entering the numbers from all locations and classification manually. Not only were they thrilled to be released from this task, but the end result is much better, too.”

Andersen-Aagaard appreciates the breadth and depth of the AI solution’s approach. “One of the best things today is when an input provider is puzzled by the machine input and asks our controllers why the AI would forecast a certain figure contrary to the knowledge of the input provider,” he adds. “When they examine it further — they find out, more often than not, that the AI is correct, because it takes all the factors and historical data into account. This makes me smile,” he says, “because then I know definitively that what we’ve created provides improved quality to our company and that we’ve saved money and time in the process.”

Andersen-Aagaard explains that the company wanted to develop a front-end and a back-end. “Our front-end is user input and confirmation, where users relate to fully and semi-fully exploded machine generated forecasts, while our back-end contains adjustable parameters that can be manually tuned when we know that parameters will change,” he says.

Ettrup explains the solution’s use of external data: “We’ve been using a lot of weather data from the past three years. Then we use that to analyze what we call a normal weather period, to generate a normal year sale per customer, per area. We gave them a lot of back-end parameters so Nukissiorfiit could control numerous company-specific aspects, such as general vacancy, project completion rate, oil per kilowatt, and much, much more.”

By migrating to the rolling forecast solution provided by CogniTech, Nukissiorfiit has reduced the number of people needed to make budget projections and increased the accuracy of those projections through increased frequency and the use of AI. The solution has improved the utility’s flexibility in adapting its planning to changing variables such as the weather and fostered greater confidence with Greenland’s government in greenlighting new and ambitious projects.

“We were able to go from 70 different input providers—employees who are involved in the process of developing our budget—down to nine people. So that’s quite a reduction,” says Andersen-Aagaard.

“And it doesn’t stop there,” he adds. “It’s actually also the amount of time that these people use it. So I would say that these nine people now spend less time than they did before. And also the 60-plus people who are not using tools anymore for providing input into the forecast, aren’t spending any time at all on it.”

Andersen-Aagaard is quick to note that while considerably fewer people work on the forecasts, every person in his company receives this information. “They get their insights both from a management point of view but also to some extent on the production data from the IBM Cognos Analytics platform,” he says.

Overall, its new Planning Analytics and Cognos Analytics solution combined with intelligent machine learning forecasts has allowed Nukissiorfiit to adopt a more efficient way of operating. The company can now use the insights to set thresholds and be alerted if the forecasts are outside of the ranges; it can also override the alerts based on experience or additional information. The bottom line is the company is more agile and its financial planning is more accurate.

“Claus [Andersen-Aagaard] is very keen to have forecasts every month to be sure the P&L and cash flow is in control. And they will be more agile to adopt changes in consumption and changes in whether a certain project is delayed or not,” says Moeller Madsen.

“Saving time has been a huge factor and benefit for us,” says Andersen-Aagaard. “Going from 70 input providers to just nine has reduced the time that we are spending on this task. And, we’ve expanded how many times we actually do this exercise—each and every month now we get a new forecast based on the latest information. The old way, we’d probably be spending 5,000–10,000 hours to do this.”

Andersen-Aagaard reports that another gain is much more precise forecasting. The company now has the flexibility it sought to change plans when new information is available and understand the consequences of doing so.

Andersen-Aagaard adds that the user experience has been tremendously improved. He reports that employees are more interested in the financial consequences of the decisions they’re making and that the output quality of the reports that they’re provided automatically has also improved drastically.

With the planning platform and machine learning in place, Nukissiorfiit is looking to the future with confidence. “I think it’s worth saying that whenever we do a big project such as the big hydroelectric plant project, you have to have confidence in us as a company,” says Andersen-Aagaard. “Greenland’s government must make sure that the money they let us invest on behalf of the county is taken care of responsibly.”

As far as the future goes, Andersen-Aagaard says the company is looking to integrate the platform with Internet of Things (IoT) sensors at its plants and built into the meters of every home in Greenland. “We’re looking into if IoT can play a bigger role so we can get data on a more frequent basis. IoT is a cost-effective platform where you can transfer a lot of data at much lower costs.”

Nukissiorfiit is also looking into exporting Greenland’s water to other countries, through working with bottling companies and bulk carriers. Plus, other Arctic countries such as Canada are very interested in seeing how they can use Nukissiorfiit’s successful utility pricing structure to reduce their own energy prices.

Nukissiorfiit (link resides outside of ibm.com) is a utility company owned by Greenland’s self-government. It is responsible for producing and supplying electricity, water and heat to the country’s consumers in 17 cities and 53 settlements. Nukissiorfiit’s energy supply is 72% based on renewable energy sources. Its goal is to use renewable energy sources wherever possible and to ensure that everyone has access to clean drinking water. The company has 405 employees, of which 90 are located at its head office in Nuuk.

IBM Business Partner CogniTech (link resides outside of ibm.com) is the only Danish IBM Gold Partner in Analytics. Founded in 2019 from a buyout of its previous iteration, the consulting organization focuses on IBM as its primary technology partner. CogniTech can create unified integrated solutions for its more than 100 customers, including everything from data integration to business intelligence to budgeting, forecasting and financial consolidation integrated with AI and machine learning. The company has more than 40 planning analytics customers.